在關鍵性零組件的製造過程中，微小瑕疵的存在是不可容忍的，因為這些瑕疵可能會削弱設計的原始強度並對整個機器系統造成嚴重危害。以飛機發動機葉片為例，應力往往會在瑕疵處集中，導致葉片斷裂，進而使飛機在空中失去推進動力。目前這類的表面瑕疵檢測主要依賴於耗時的人工目視檢查，隨著少子化的趨勢和深度學習技術的快速發展，自動化檢測已逐漸成為製造業中一個既可行又必要的發展方向。然而，實現100%的檢測準確度仍是一大挑戰，任何未檢出的瑕疵品都可能造成嚴重的後果，因此目前仍須依靠人工檢查作為保障步驟。
而本研究旨在通過自動化檢測技術預先識別並排除瑕疵品，以減輕人工目視檢查的負擔。研究重點在於提高召回率(Recall)，即降低正常產品被誤判的機率，同時提升特異度(Specificity)，即減少需要人工檢查的瑕疵產品比例。
本研究透過訓練單一樣本的自編碼器模型(Autoencoder, AE)作為主要檢測模型，此模型適合應用於擁有大量正常樣本與少量瑕疵樣本的情境中。並額外運用了少量的瑕疵樣本來訓練自編碼器，發現到加入瑕疵樣本來訓練自編碼器並混合使用，更能發現到正常樣本資料的集中模式。最後再通過數學分類模型來混合不同自編碼器的輸出結果來設定適當閾值，相比於傳統的自編碼器採最大閾值方法，本研究的召回率從99%提高至100%，而特異度則從35.5%提升至48.5%，顯著提高了檢測性能。

In the manufacturing of critical components, minor defects are intolerable as they can weaken the original strength of the design and cause severe damage to the entire system. For example, defects in aircraft engine blades can concentrate stress, leading to fractures and loss of propulsion. Currently, detecting such defects relies on time-consuming manual inspections. With fewer workers and advancements in deep learning, automated inspection is becoming essential. Achieving 100% accuracy is challenging, so manual checks are still needed as a safeguard.
This study aims to reduce the burden of manual inspections by using automated techniques to identify and eliminate defects. The focus is on improving recall rates to reduce false negatives and enhancing specificity to lower the need for manual inspections.
The study uses an autoencoder (AE) model trained on single samples, suitable for scenarios with many normal and few defective samples. Including some defective samples in training improves the detection of normal patterns. By combining outputs from different autoencoders with a mathematical model, the recall rate improved from 99% to 100%, and specificity from 35.5% to 48.5%, significantly enhancing detection performance.

英文:
Antoniou, A., Storkey, A., & Edwards, H. (2017). Data augmentation generative adversarial networks. arXiv preprint arXiv:1711.04340.
Arjovsky, M., & Bottou, L. (2017). Towards principled methods for training generative adversarial networks. arXiv preprint arXiv:1701.04862.
Arjovsky, M., Chintala, S., & Bottou, L. (2017). Wasserstein generative adversarial networks. In Proceedings of the 34th International Conference on Machine Learning, 214-223. PMLR.
Baur, C., Denner, S., Wiestler, B., Navab, N., & Albarqouni, S. (2021). Autoencoders for unsupervised anomaly segmentation in brain MR images: a comparative study. Medical Image Analysis, 69, 101952.
Bergmann, P., Batzner, K., Fauser, M., Sattlegger, D., & Steger, C. (2021). The MVTec anomaly detection dataset: a comprehensive real-world dataset for unsupervised anomaly detection. International Journal of Computer Vision, 129(4), 1038-1059.
Bergmann, P., Löwe, S., Fauser, M., Sattlegger, D., & Steger, C. (2018). Improving unsupervised defect segmentation by applying structural similarity to autoencoders. arXiv preprint arXiv:1807.02011.
Breiman, L. (2001). Random forests. Machine learning, 45, 5-32.
Chen, T., & Guestrin, C. (2016). Xgboost: A scalable tree boosting system. In Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, 785-794.
Deng, J., Dong, W., Socher, R., Li, L. J., Li, K., & Fei-Fei, L. (2009). ImageNet: A large-scale hierarchical image database. In 2009 IEEE Conference on Computer Vision and Pattern Recognition, 248-255. IEEE.
Fawcett, T. (2006). An introduction to ROC analysis. Pattern Recognition Letters, 27(8), 861-874.
Gatys, L. A., Ecker, A. S., & Bethge, M. (2015). A neural algorithm of artistic style. arXiv preprint arXiv:1508.06576.
Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., & Bengio, Y. (2014). Generative adversarial nets. In Proceedings of the 27th International Conference on Neural Information Processing Systems, 2, 2672-2680.
Hensman, P., & Masko, D. (2015). The impact of imbalanced training data for convolutional neural networks (Bachelor's thesis). KTH Royal Institute of Technology.
Jarvis, J. F. (1979). Visual inspection automation. In COMPSAC 79. Proceedings. Computer Software and The IEEE Computer Society's Third International Applications Conference, 1979, 251-255. IEEE.
Jiang, X., Xie, G., Wang, J., Liu, Y., Wang, C., Zheng, F., & Jin, Y. (2022). A survey of visual sensory anomaly detection. arXiv preprint arXiv:2202.07006.
Khalifa, N. E., Loey, M., & Mirjalili, S. (2022). A comprehensive survey of recent trends in deep learning for digital images augmentation. Artificial Intelligence Review, 55, 2351-2377.
Kim, J., Jeong, K., Choi, H., & Seo, K. (2020). GAN-based anomaly detection in imbalance problems. In Proceedings of Computer Vision–ECCV 2020 Workshops, 128-145.
Kingma, D. P., & Welling, M. (2013). Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114.
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. Advances in Neural Information Processing systems, 25.
Lei, C., Hu, B., Wang, D., Zhang, S., & Chen, Z. (2019). A preliminary study on data augmentation of deep learning for image classification. In Proceedings of the 11th Asia-Pacific Symposium on Internetware, 1-6.
O’Mahony, N., Campbell, S., Carvalho, A., Harapanahalli, S., Hernandez, G. V., Krpalkova, L., Riordan, D & Walsh, J. (2020). Deep learning vs. traditional computer vision. In Advances in Computer Vision: Proceedings of the 2019 Computer Vision Conference, 1(1), 128-144.
Perkins. (1978). A model-based vision system for industrial parts. IEEE Transactions on Computers, 100(2), 126-143.
Radford, A., Metz, L., & Chintala, S. (2015). Unsupervised representation learning with deep convolutional generative adversarial networks. arXiv preprint arXiv:1511.06434.
Schlegl, T., Seeböck, P., Waldstein, S. M., Langs, G., & Schmidt-Erfurth, U. (2019). f-AnoGAN: Fast unsupervised anomaly detection with generative adversarial networks. Medical Image Analysis, 54, 30-44.
Schlegl, T., Seeböck, P., Waldstein, S. M., Schmidt-Erfurth, U., & Langs, G. (2017). Unsupervised anomaly detection with generative adversarial networks to guide marker discovery. In International Conference on Information Processing in Medical Imaging, 146-157.
Schmarje, L., Santarossa, M., Schröder, S. M., & Koch, R. (2021). A survey on semi-, self-and unsupervised learning for image classification. IEEE Access, 9, 82146-82168.
Shorten, C., & Khoshgoftaar, T. M. (2019). A survey on image data augmentation for deep learning. Journal of Big Data, 6(1), 1-48.
Springenberg, J. T., & Riedmiller, M. (2013). Improving deep neural networks with probabilistic maxout units. arXiv preprint arXiv:1312.6116.
Stoltzfus, J. C. (2011). Logistic regression: a brief primer. Academic emergency medicine, 18(10), 1099-1104.
Strelcenia, E., & Prakoonwit, S. (2023). Effective feature engineering and classification of breast cancer diagnosis: a comparative study. BioMedInformatics, 3(3), 616-631.
Wang, Z., Bovik, A. C., Sheikh, H. R., & Simoncelli, E. P. (2004). Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Processing, 13(4), 600-612.
Xia, X., Pan, X., Li, N., He, X., Ma, L., Zhang, X., & Ding, N. (2022). GAN-based anomaly detection: A review. Neurocomputing, 493, 497-535.
Xiao, Q., Liu, B., Li, Z., Ni, W., Yang, Z., & Li, L. (2021). Progressive data augmentation method for remote sensing ship image classification based on imaging simulation system and neural style transfer. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 14, 9176-9186.
Yang, J., Xu, R., Qi, Z., & Shi, Y. (2021). Visual anomaly detection for images: A survey. arXiv preprint arXiv:2109.13157.
Zhang, H. W., Huang, D. L., Wang, Y. R., Zhong, H. S., & Pang, H. W. (2024). CT radiomics based on different machine learning models for classifying gross tumor volume and normal liver tissue in hepatocellular carcinoma. Cancer Imaging, 24(1), 20.

中文:
行政院飛航安全委員會（2010）。飛航事故調查報告ASC-AOR-10-08-001。中華民國行政院。https://www.ttsb.gov.tw/media/3512/asc-aor-10-08-001.pdf
國家發展委員會（2023）。人口推估統計圖表彙編-2022年版。中華民國行政院。https://ws.ndc.gov.tw/Download.ashx?u=LzAwMS9hZG1pbmlzdHJhdG9yLzEwL3JlbGZpbGUvMC8xMTI4LzBkOTZlZjBjLTkwMTUtNDc0ZS05MzBkLTY1MWVkZGFhZjQ0OS5wZGY%3d&n=5Lq65Y%2bj5o6o5Lyw57Wx6KiI5ZyW6KGo5b2Z57eoLnBkZg%3d%3d&icon=.pdf

網路資料:
A03ki. (2022). f-AnoGAN [Code]. https://github.com/A03ki/f-AnoGAN